Analysis of the influence of ENSO phenomena on wave

Transcription

Analysis of the influence of ENSO phenomena on wave
Revista de Gestão Costeira Integrada / Journal of Integrated Coastal Zone Management, 15(3):353-370 (2015)
http://www.aprh.pt/rgci/pdf/rgci-570_Pereira.pdf
|
DOI:10.5894/rgci570
Analysis of the influence of ENSO phenomena
on wave climate on the central coastal zone of Rio de Janeiro (Brazil)
@,
Nair Emmanuela da Silveira Pereira@, a; Leonardo Azevedo Klumb-Oliveirab
Abstract
This paper evaluates the influence of the El Niño – Southern Oscillation on the wave climate variability in the central region of
Rio de Janeiro’s coastal zone. The regional climate of the area was characterized using the WAVEWATCH III wave data
model, obtained from NOAA, organized as a 35 year time series (1979-2013). These data were validated for the study area and
a characterization and analysis performed by focusing on years with occurrence of strong El Niño/La Niña events. The correlation between the interannual variability of significant wave height and Oceanic Niño Index showed a slight reduction in significant wave height during El Niño years and the opposite pattern during La Niña years, with a lag of four months. This decrease could be attributed to the intensification of the South Atlantic High with a corresponding increase in the occurrence of
subtropical jets during periods of El Niño. This weather change causes the blocking of cold fronts in the southern region of
Brazil and the consequent reduction in the percentage of waves from the south along the southeast coast.
Keywords: climate variability, wave parameters, interannual time scale, seasonality, regional climatology
Resumo
Análise da influência de fenômenos ENOS no clima de ondas da porção central da zona costeira do Rio de Janeiro
(Brasil)
Este artigo busca avaliar a influência do fenômeno El Niño - Oscilação Sul na variabilidade do clima de ondas da porção central da zona costeira do Rio de Janeiro. Para tanto, a climatologia regional da área foi caracterizada a partir de dados de ondas
do modelo WAVEWATCH III disponibilizado pela NOAA, organizados numa série temporal de 35 anos (1979 - 2013). Esses
dados foram validados para a área de estudo e, realizada sua caracterização e análise com enfoque em anos de acentuada ocorrência de eventos de El Niño/La Ninã. A correlação da variabilidade interanual da série de altura significativa de ondas e Índice de Niño Oceânico mostrou padrão de leve redução das alturas significativas em anos de El Niño e, o inverso para anos de
La Niña, com defasagem temporal de quatro meses na região. Essa redução na altura significativa das ondas pode ser atribuída
à intensificação da Alta Subtropical do Atlântico Sul e aumento na ocorrência de jatos subtropicais nos períodos de El Niño.
Essa alteração no padrão meteorológico causa o bloqueio de frentes frias na região Sul do Brasil e consequente redução da
porcentagem de ondas de sul na região sudeste.
Palavras-chave: variação climática, parâmetros de onda, escala interanual, sazonalidade, climatologia regional
@
a
b
Corresponding author, to whom correspondence should be addressed:
MAG - Mar, Ambiente e Geologia – Serviços. Rua Visconde de Inhaúma, 37 - 21º andar, Centro, CEP: 20091-007, Rio de Janeiro, RJ,
Brazil. e-mail: Pereira <[email protected]>
Universidade Federal do Rio de Janeiro - Programa de Pós-Graduação em Geografia. Avenida Athos da Silveira Ramos, 274 - Prédio do
CCMN - Bloco I - Cidade Universitária, CEP: 21.941-916, Rio de Janeiro, RJ, Brazil. e-mail: Klumb-Oliveira <[email protected]>
* Submission: 21 DEC 2014; Peer review: 17 JAN 2015; Revised: 9 MAR 2015; Accepted: 28 MAR 2015; Available on-line: 30 MAR 2015
This article contains supporting information online at http://www.aprh.pt/rgci/pdf/rgci-570_Pereira_Supporting-Information.pdf
Pereira & Klumb-Oliveira (2015)
1. Introdução
The El Niño – Southern Oscillation (ENSO) is the
dominant mode in ocean-atmosphere coupling variability on interannual time scales (Trenberth & Stepaniak,
2001). The El Niño is characterized as an abnormal
warming in the sea surface that occurs during some
years in the coastal zone of Peru and Colombia, changing the pattern of the climate systems on both local and
regional levels (Trenberth, 1997). The La Niña is the
opposite phase of the El Niño, characterized as an abnormal cooling in the surface water.
Although the El Niño is a regional phenomenon in the
Pacific region, the climatic anomalies associated with
ENSO are nearly global in extent and have a highly
persistent nature (Kousky et al., 1984). The ENSO atmospheric component, the “Southern Oscillation”, corresponds to a zonal balance in the large scale air masses
with a direct response in the variability of the atmospheric pressure (Aragão, 1998).
Knowledge of environmental conditions has fundamental importance for oceanic engineering (Souza &
Ribeiro, 1988). As an example, beach morphology is
dependent on the combined actions of local environmental conditions, sediment type and the previous wave
behavior (Komar, 1976; Short, 1999; Muehe, 2011).
Knowledge of the regional wave climate, in conjunction
with that of morphological processes in the coastal
zone, strengthens coastal management studies and the
implementation of engineering structures at the coast
(Pianca et al., 2010).
Comprehending the relationship between ENSO events
and changes in environmental parameters has numerous
applications, such as the evaluation of the energetic
potential of winds and waves in the coastal zone and in
coastal management to support public policies. In a
study of the energy potential of wind in the state of
Ceará (Brazil), Araújo Júnior et al. (2014) verified that
conditions to generate wind energy were better during
the 1997-1998 El Niño period than during the La Niña
event of 1998-1999, principally on the coast.
Impact studies of El Niño/La Niña events on regional
climates can provide improvements to climate prediction techniques to support public policies. Changnon
(1999) studied the impact of the 1997-1998 El Niño
event on the economy and loss of life in the United
States, using the state of California as an example. An
analysis of ENSO predictions and ENSO’s economic
and social effects led the government of California to
conduct impact reduction efforts, which decreased subsequent economic and human losses.
As yet, Brazil is still lagging studies that identify wave
patterns during El Niño periods, and, as a consequence,
also lacks the public policies regarding coastal management and coastal hazards that relate to those events.
Accordingly, this work aims to characterize the regional
wave climate of the central coastal zone of Rio de Janeiro through the analysis of a global wave model and
the evaluation of apparent changes in the patterns of
this data on an interannual time scale, focusing on several years during which El Niño and La Niña events
occurred.
2. Methodology
2.1. Study area
The central area of the coastal zone of Rio de Janeiro
used in this study is comprised of the waters from the
continental slope to the coastal waters near the shoreline (Figure 1).
The wave climate for this region is that of predominantly “good weather” conditions with NE winds and a
presence of storm waves mainly from the South Atlantic High (SAH) associated with frontal systems from
the east-southeast (Silva et al., 2009).
Observations indicate that the mean wave height varies
between 1.6 to 2.0 m during the “good weather” conditions. Waves with height greater than 3.0 m frequently
come from the S and SW quadrants (Bastos & Silva,
2000). Waves from the NE and E are low energy but
the most frequent, based on a comparison of the wave
energy distribution by incidence direction (Muehe &
Corrêa, 1989). According to Diretoria de Hidrografia e
Navegação (DHN/Navy of Brazil), the tidal variation
ranges from 1.3 m at spring tides to 0.3 m at neap tides.
2.2. Data set
To determine the wave climate during El Niño/La Niña
event along the Rio de Janeiro coast, simulation data
was obtained from the WAVEWATCH III wave model,
available from the National Oceanic and Atmospheric
Administration (NOAA), with a global spatial resolution of 0.5º and 3 h temporal resolution (NOAA
WAVEWATCH III - NWW3).
Two versions of the model dataset were used. The first
was a reanalysis of historical waves hindcast using
WAVEWATCH III, version 2.22 (Tolman, 2002) for
the period from January 2010 to December 2013. The
second was a reanalysis of waves hindcast using
WAVEWATCH III, version 3.14 (Tolman, 2009) with
input data from the Climate Forecast System Reanalysis
Reforecast (CFSRR) wind reanalysis (Spindler et al.,
2011), obtained from the National Center for Environmental Prediction (NCEP) from January 1979 to December 2009 (Chawla et al., 2012, 2011).
From this data, the variables used were significant wave
height, peak period and peak direction (ranging from 0
to 360º, where 0 represents waves from the north)
which were extracted for the total period from January
1979 to December 2013 (35 years). To represent the
354
Revista de Gestão Costeira Integrada / Journal of Integrated Coastal Zone Management, 15(3):353-370 (2015)
Figure 1 - Study area used in the wave climate analysis, shown using bathymetry from TOPEX/POSEIDON
(Becker et al. 2009) near Cabo Frio, on coastal region of Rio de Janeiro, Brazil. The black dots represent the
spatial resolution of NOAA WAVEWATCH III model data.
Figura 1 - Área de estudo utilizada na análise do clima de ondas, demonstrada por batimetria TOPEX/POSEIDON (Becker et al. 2009) próximo a Cabo Frio, na região costeira do Rio de Janeiro, Brasil. Os
pontos em preto representam a resolução espacial dos dados do modelo NOAA WAVEWATCH III.
central coastal region of Rio de Janeiro, near the city of
Cabo Frio, the area was defined as between 22.5 - 23.5º S
and 42.5 - 41.5º W (Figure 1).
The Oceanic Niño Index (ONI) series used are available
from NOAA’s Climate Prediction Center (Figure 2).
This index was calculated as a 3 month moving average
of the temperature anomaly over the period under
analysis using the Extended Reconstructed Sea Surface
Temperature - ERSST.v3b dataset (Smith et al., 2008)
for the Niño 3.4 region (5º N – 5º S, 120º W – 170º W).
According to Larkin & Harrison (2005), NOAA defines
El Niño events as periods of three consecutive months
with an ONI that exceeds 0.5ºC, while La Niña events
are identified for an ONI below -0.5ºC.
From this index, we selected the years 1997 and 1998,
which represented the beginning and the end, respectively, of the strongest El Niño period of the 20th Century (Wolter & Timlin, 1998). We also selected the year
2000, a representative year within a long period of La
Niña (from 1998 to 2001) that occurred after the 19971998 El Niño, classified as a moderate to intense event
(Shabbar & Yu, 2009). For a more significant representation of the mean pattern in El Niño/La Niña years, the
years with major temporal persistence of the phenomena were chosen (1983, 1987, 1992, 1997 and 2003 for
El Niño years and 1988, 2000 and 2008 for La Niña
years).
For the correct application of the model’s data, adequate validation is necessary. Data validation of
NWW3 was performed using a statistical comparison to
data from a stationary meteo-oceanographic buoy using
a period from July 17 to December 31, 2013. This buoy
from the Instituto de Estudos do Mar Almirante Paulo
Moreira (IEAPM/Navy of Brazil) is part of the SIODOC project (Sistema Integrado de Obtenção de Dados
Oceanográficos) and is located at 22º 59.62’ S – 42º
11.65’ W. The resolution of the data was in 1 hour time
intervals and is available at http://metocean.fugrogeos.
com/marinha/ .
2.3. Statistical treatment and model validation
For many applications, a time series can be considered
as a linear combination of periodic or quasi-periodic
components (with fixed amplitudes and phases) on
which a tendency and high frequency noise are superposed (Emery & Thomson, 2001). Fourier analysis
355
Pereira & Klumb-Oliveira (2015)
Figure 2 - Temporal variability of the Oceanic Niño Index for the period from Jan 1979 to Dec 2013, with the occurrences of
El Niño periods presented in red and La Niña events in blue. This monthly index considers the 3 month averages of the
ERSST.v3b sea surface temperature anomaly Data source: http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ ensoyears.shtml.
Figura 2 - Variabilidade temporal do Índice de Niño Oceânico para o período de Janeiro de 1979 a Dezembro de 2013, com a
ocorrência de períodos de El Niño representados em vermelho e, os eventos de La Niña em azul. Esse índice mensal considera a média de 3 meses da anomalia de temperatura do mar em superfície do ERSST.v3b Fonte dos dados
http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ensoyears.shtml.
thereby considers as a basic premise that the time series
can be reproduced as a sum of wave signals (sines and
cosines) with the mean value of the series. The time
series was decomposed to observe the individual signature of each wave signal. To calculate the significance
of the power spectrum’s peaks, the red noise was computed based on a first order autoregression process
(Zhang & Moore, 2011). Peaks in the time series’
power spectrum that are above the red noise’s Fourier
spectrum have a 95% confidence level.
For the statistical analysis, the time scale of the buoy
data was reduced to obtain 3 hour means. A spectral
filter was used to identify and remove high frequency
signals. The high frequency noise can be related to a
series of factors, such as characteristics of the equipment used. The significant wave height and peak period
were chosen for analysis from the buoy data because
the peak direction data was not available for the buoy
dataset. The NWW3 model data used to compare with
the buoy data uses the point 23º S – 42º W, which was
the closest to the point where the buoy was anchored.
The mean and standard deviation were calculated for
each variable over the analysis period. The correlation
coefficient is used to determine how well two (or more)
variables co-vary in time (or space) (Emery & Thomson, 2001). Considering two distinct series varying in
time, the linear correlation coefficient can be calculated
considering its displacement in time, referred to as cross
correlation.
This coefficient suggests the linear dependency of the
compared data because the values indicate their degree
of dispersion around the adjustment function (linear
regression). It functions as a quantifier of the intensity
and direction of this linear relationship (Chen et al.,
2013). The values of this parameter oscillate between 1
and -1, where values near zero indicate that the variables are not related. High positive values indicate that
the variables exhibit similar behavior, while high negative values indicate an inverse relationship in the behavior of the variables.
Considering an analysis focused on interannual variability, a statistical treatment was applied to the time
series of the wave parameters to remove the seasonal
signature and calculate cross correlations between the
wave parameters and the ONI. The “climatology”
method was used, which consisted of subtracting the
monthly climatology from the 35 years of the series that
had previously been reduced to monthly time steps. A 3
months running mean was then applied. This is the
same methodology used to obtain the ONI data (Douglass, 2011). The objective of this filtering was to remove characteristic oscillations due to seasonality and
other events with high frequency such as frontal system
passages.
Verification that both variables are related can be obtained from the empirical evidence that there is a correspondence in the patterns of those variables higher than
expected than if using random data (Orcutt & James,
1948). To test the significance of the correlations between wave parameters and the ONI, random series
were generated from each of the original series. These
were correlated, forming a population of correlations
comprising a total of 10,000 samples.
The probability of a random variable showing a determined value is represented by the Cumulative Distribution Function (CDF). A non-parametric estimate was
356
Revista de Gestão Costeira Integrada / Journal of Integrated Coastal Zone Management, 15(3):353-370 (2015)
used to adjust the CDF by determining an empirical
(ECDF), thus without use of a theoretical model with a
specific distribution.
In this study, the Kaplan-Meier estimator was used to
estimate the CDF (Kaplan & Meier, 1958; Lawless,
2002) as an empirical function, considering a test for
full population (N = 10,000) and another for a subset of
that population (N = 200). For comparison with the
empirical method, an adjustment of the theoretical CDF
of the resultant population distribution using a normal
function (Gaussian) was selected.
For the validation of NWW3 data with the buoy data
and the comparison between theoretical and empirical
CDFs, the mean absolute error (MAE) and relative
(RMAE) were also calculated. The RMAE is the MAE
normalized using the average values of the buoy data to
obtain the proportion of the model error to the mean
value observed in nature. The MAE is one of the best
parameter choices to quantify the average error, using
the magnitude of the absolute difference between the
modeled and observed data at each point of time series
(Willmott, 1982; Willmott & Matsuura, 2005). For this
parameter, zero represents the perfect adjustment between modeled and observed data.
As part of the NWW3 validation, a parameter proposed
by Willmott (1981) was also calculated. A parameter
value of 1 is a perfect fit between the results obtained
by the model and those observed, while 0 represents a
complete mismatch. Although the correlation indicates
the interdependence of the series (the degree of dispersion of the series considering a linear adjustment), the
Willmott parameter is indicative of the similarity in the
compared data, related to the remoteness of the estimated values observed in both series (Chen et al.,
2013).
3. Results and discussion
3.1. NWW3 data validation
The following image (Figure 3) shows the significant
wave height and peak period data of NWW3 contrasted
with buoy data acquired near Arraial do Cabo, before
and after receiving spectral treatment. If the high frequency spectrum was removed from the data, it did not
change the overall behavior of the curve, but did eliminate the high frequency noise in the original data. Thus,
there is a reduction of the standard deviation of the
filtered data values from the original, and a slight attenuation of extreme events.
It is remarkable that the modeled data show an overall
behavior similar to data found in the environment. The
linear correlation index calculated for the buoy data
after spectral filtering was approximately 0.80 for the
Figure 3 - Comparison between the original wave data from the buoy, where can be observed the same data before and after
filtering and, the NWW3 data. The statistical parameters referring to the comparison between buoy data after filtering and
NWW3 data are presented for: (a) significant wave height and; (b) peak period.
Figura 3 - Comparação entre os dados de onda originais provenientes da boia, onde podem-se observar os mesmos dados
antes e depois da filtragem e, o dado do NWW3. Os parâmetros estatísticos referentes à comparação entre os dados de
boia após filtragem e os dados o NWW3 são apresentados para: (a) altura significativa da onda e; (b) período de pico.
357
Pereira & Klumb-Oliveira (2015)
significant wave height and 0.75 for the peak period.
The Willmott parameter values were 0.95 for significant wave height and 0.92 for the peak period.
In this context, no reference values have been established for the Wilmott parameter, but it is generally
accepted that values around 0.5 indicate that the model
reproduces about half of the observed variance (Hetland
& Dimarco, 2012). Chen et al. (2013) consider an indication of good accuracy to be correlation values greater
than 0.77 and the Willmott parameter above 0.70. The
high values for both parameters indicate that in addition
to showing a similar oscillation pattern, the data sets
show a strong fit, which demonstrates that the difference between each pair of points for this series is similar for the entire set of series.
On first examination, observing the average values of
the series for both the model and buoy data, a notable
overestimation by the model is evident. The mean modeled significant wave height is approximately 0.47 m
higher in comparison to the values given by the buoy.
The same occurs for peak period, but with a smaller
difference, with an average of 0.16 s greater for the
model than for the buoy. For either the significant wave
height or the peak period, the differences between the
sampled and modeled mean values lie within the standard deviation range.
The average error, which quantifies the mean difference
between the model values in comparison to the observed data from the buoy, was approximately 0.51 m
for significant wave height (35.69% error), while for
the peak period this value was 0.99 s (10.26% error). A
comparison between observed data and NWW3 near the
Florianópolis region by Pianca et al. (2010) confirms
the similarity between model and environment, showing
the same pattern of model’s overestimation for the significant wave height data.
Hanson et al. (2009) reviewing NWW3 performance in
the Pacific Ocean also found a remarkable correlation
with observational data, greater than 0.78 for wave
height and 0.88 for the period. That study found a
maximum error of 0.5 m in wave height that is associated with the use of the hindcast altimetry data, which
tends to overestimate wave height values. Rogers et al.
(2012) found a 10% overestimation in the referred data.
For both variables in the NWW3 data we can observe
that the first two weeks of November presented a constant value, differently of the buoy data. This can represent many types of errors, probably related to the data
assimilated by the model in the reanalysis process. The
period occurs after high values of significant wave
height and peak period, which can suggest the approximation of a cold front passage in the study area both in
the beginning and the middle of the period. These consecutive passages of cold fronts can promote cloud
cover persistence, which prevents data collection via
remote sensing.
The validation authenticates the use of these data for
environmental studies. However, although validated for
the North and Tropical Atlantic, models on a global
scale such as this need further comparison with data
collected in the vicinity of the south and southeast Brazilian Continental Shelf. Data sets such as NWW3 formerly had a significant discrepancy in the Southern
Hemisphere, particularly as a consequence of few
measurements in situ in these regions (Caires et al.,
2004). This type of validation would improve models
with global resolution in the South Atlantic region.
3.2. Wave climate characterization
Starting from the analysis of time series for wave parameters in the region (see Supporting Information I), it
can be seen that the significant wave height for the
study area has an average of approximately
1.80 ± 0.51 m, with marked seasonality that results in an
average oscillation limited by the standard deviation.
This same seasonal pattern is observed for both the
peak direction (147.42 ± 43.52º) and peak period
(9.80 ± 2.25 s).
In this context, the months of May and September exhibited a pattern of higher significant heights predominantly from the South quadrant, with long periods
(around 13 s). In contrast, the austral summer is characterized by lower significant wave heights predominantly from the East with shorter periods (around 7.5 s).
Figures 4a, 4b and 4c show that significant wave height
varies between 1 and 4 m for the majority of the values,
with the highest concentration of values around the
average. The peak period shows the same pattern, with
samples concentrated between 5 and 15 s and a predominance of samples around the average value. The
peak direction ranges from 50° to 220° and presents a
bimodal pattern, with one mode centered at 90° and
another around 180°.
The monthly climatology of wave parameters (Figures
4d, 4e and 4f) show that the maximum average values
for significant wave height occur at the end of the austral winter (with average values of approximately
2.0 m) and the minimum in late summer (with average
values of 1.6 m). Unlike significant wave height, the
direction and peak period have maximum average values in late autumn (with average values around 10.5 s SSE) and their minimum in the middle of summer (with
average values around 8.5 S-SE). The maximum values
for these variables preceded the maximum values of
significant wave height.
Surface gravity waves can be categorized into two
types. The first type, "wind sea" waves, have periods of
less than 10 s with an irregular appearance and short
358
Revista de Gestão Costeira Integrada / Journal of Integrated Coastal Zone Management, 15(3):353-370 (2015)
Figure 4 - Histogram referencing the time series data of significant wave height (a), peak direction (b) and peak period (c) for
the period from Jan 1979 to Dec 2013 and considering a sample size of 102,272 points. The monthly climatology for the
same period is shown for the same parameters in (d), (e) and (f), respectively.
Figura 4 - Histograma referente aos dados de série temporal de altura significativa da onda (a), direção de pico (b) e período
de pico (c) para o período de Janeiro de 1979 a Dezembro de 2013 e, considerando um tamanho amostral de 102272 pontos. A climatologia mensal para o mesmo período é apresentada para os mesmos parâmetros em (d), (e) e (f), respectivamente.
wave lengths, and are generated by local winds. The
second type are "swell" waves, which have periods
between 10 and 20 s, characteristic long wave lengths
and an almost sinusoidal form, present far from their
generation area (Holthuijsen, 2007; Laing, 1998).
The wave regime in this region is controlled by the
South Atlantic High and the passage of cold fronts
(Pianca et al., 2010). Thus, with respect to the region
under consideration, the first mode (NE-SE) is characterized as waves associated with the South Atlantic
Anticyclone as a result of the downward arm of the
Hadley cell (Campos, 2009) which promotes moderate
winds. The associated "good weather" waves have a
low peak period (up to 7.5 s). The second mode (S-SW)
is associated with the migration of polar air masses
forming cold fronts in the region; these waves present
peak periods between 11 and 15 s and carry more energy, noted by Muehe & Corrêa (1989), Pianca et al.
(2010) and Souza & Ribeiro (1988).
The directional histograms for significant wave height
and peak period (Figure 5) distinctly show the same
general pattern. The highest concentration of waves is
from the quadrant between the NE and SW. The highest
significant wave height values are concentrated between the SE and SW, with predominant periods longer
than 10 s. The waves with shortest periods (less than
10 s) originated from the direction between NE and SE
were the minor percentage of the total These two patterns of waves represent the two distinct types of wave
characterized by Souza & Ribeiro (1988). As seen in
these data, Bastos & Silva (2000) similarly observed
that for the region, the occurrence of waves higher than
3.0 m most often originates in the S and SW.
Figure 6 shows climatological directional histograms
observed for the four seasons of the year, in which all
seasons maintain the mean pattern of having greater
significant wave heights and peak periods from the S.
In the austral summer, the waves appear to be distrib-
359
Pereira & Klumb-Oliveira (2015)
greatest percentage of extreme events of NE waves in
the Campos Basin region occurred during the austral
summer and had a lower incidence in the fall.
3.3. Interannual variability and El Niño interference
in regional climatology
Figure 5 - Directional histograms of wave climatology for the
period from Jan 1979 to Dec 2013 in 15° intervals, showing: (a) the significant wave height and; (b) the respective
peak period.
Figura 5 - Histogramas direcionais da climatologia de ondas
para o período de Janeiro de 1979 a Dezembro de 2013,
em intervalos de 15o, mostrando em: (a) altura significativa de onda e; (b) respectivo período de pico.
uted in a uniform pattern from the E to S quadrants,
while in autumn, a higher concentration of wave samples originates from the S and SW, presenting a nearly
unimodal pattern. This pattern is repeated in the winter
with a greater occurrence of waves from the E. It returns to a more homogeneous pattern in the spring,
however, with a larger signature of S-SW waves and
higher rates of waves with over 1.8 m heights. This
pattern is similar to that verified by Pianca et al. (2010)
for the Brazilian southeast region.
Concurring with our findings, in a wave climatology
study from 1991 to 1995, Vale (2012) observed that the
We can observe in Figure 7, that the periods of 1 and
1.5 years are detached in the Fourier’s spectrum from
the time series of the three wave parameters under
analysis, as discussed previously. The seasonality and
other signals with high frequencies observed in this
time series masked the observation of interannual variability.
We show, in Figure 8, the frequency spectrum related to
the ONI for the period 1979 to 2013 with the same
wave data after statistical treatment to remove seasonality. The highest spectral power values refer to frequencies lower than 1 cycle by year (c.b.y). It is remarkable
that after statistical treatment to remove seasonality, the
wave data showed greater emphasis on low frequency
signals without significantly changing the default signature. However, the high frequencies were still very
prominent. This noise should have a negative influence
on the correlations between these data and the ONI.
Comparing the ONI and the significant wave height
after removal of seasonality no clear correlation is apparent between all the datasets for the three parameters,
which is confirmed by the low linear correlation values
shown (see Supporting Information II). However, according to Trenberth & Stepaniak (2001), the definition
of El Niño indices (which use anomaly temperatures in
the Pacific to identify the El Niño and La Niña periods)
do not permit the discrimination of the characteristics of
its effects in other regions; this causes difficulty in
mapping a relationship between this index and the variability of some parameters on the Brazilian coast.
Examining cross-correlations between series for parameters concerning the wave and the ONI (Figure 9),
the highest correlation coefficient was observed for the
significant wave height parameter, approximately -0.19,
with a time delay of four months. This delay suggests
that the most likely correlation occurs with a 4-month
response lag. The peak direction showed a negative
correlation of approximately -0.14, with a lag of 8
months. The peak period, however, showed a positive
correlation of 0.11 with zero lag. These correlation
values, although apparently small, are above the minimum level of significance, even when using tests with a
reduced N sample size (Figure 10). We need to consider
the signature of the high frequency in wave parameters’
data, which can reduce comparison of these two
datasets.
This data indicates that in El Niño years (positive ONI),
there would be a slight reduction in significant wave
heights in the coastal region of Rio de Janeiro, with a
360
Revista de Gestão Costeira Integrada / Journal of Integrated Coastal Zone Management, 15(3):353-370 (2015)
Figure 6 - Directional histograms of wave climatology for the period from Jan 1979 to Dec 2013, in 15° intervals. The significant wave height and peak periods, respectively, are presented for austral seasons of : (a) and (b) - summer (DJF); (c) and
(d) - autumn (MAM); (e) and (f) - winter (JJA) and ; (g) and (h) - spring (SON).
Figura 6 - Histogramas direcionais da climatologia de ondas para o período de Janeiro de 1979 a Dezembro de 2013, em
intervalos de 15º. A altura significativa das ondas e períodos de pico, respectivamente, são apresentados para as estações
do ano no hemisfério sul: (a) e (b) – verão (DJF); (c) e (d) – outono (MAM); (e) e (f) – inverno (JJA) e; (g) e (h) – primavera (SON).
361
Pereira & Klumb-Oliveira (2015)
Figure 7 - Frequency Spectrum referring to the significant wave height (a), peak direction (b) and peak period data (c) for the
period from Jan 1979 to Dec 2013. The red noise spectrum represents a 95% confidence level and the units of frequency
are presented in cycles by year (c.b.y.).
Figura 7 - Espectro de frequências referente aos dados de altura de onda significativa (a), direção de pico (b) e período de
pico (c) para o período de Janeiro de 1979 a Dezembro de 2013. O espectro do “ruído vermelho” representa num nível de
confiança de 95% e, as unidades de frequência são apresentadas em ciclos por ano (c.b.y. – sigla em inglês).
delay of 4 months from the peak of the El Niño in Pacific waters. A study by Enfield & Mayer (1997)
showed a correlation of -0.3 with a gap of 4 months
between the temperature anomalies in the surface sea
water in the South Atlantic and the Pacific surface temperature index, an index similar to the NINO3 index
that is often used in studies of ENSO cycles (Enfield &
Mayer, 1997). In the same study, the wind data preceded the response by two months, the time expected
for variations in atmospheric patterns to be reflected in
the surface of the sea.
After presuming that there was no strong correlation
between the ONI and the other wave parameters, we
chose to observe the wave pattern for each season of the
year according to the index shown in Figure 2, considering years with significant El Niño and La Niña
events. These data were compared for a more detailed
observation of differences during ENSO periods using
the climate pattern shown in the 31 year series of wave
behavior data and focusing on the differences in the
seasonality of these parameters.
The beginning of an El Niño period, 1997, was examined to compare its wave pattern (see Supporting Information III) to the wave climatology (Figure 6). In the
summer, a smaller percentage of incident waves with
more than 1.8 m in height were observed, along with a
362
Revista de Gestão Costeira Integrada / Journal of Integrated Coastal Zone Management, 15(3):353-370 (2015)
Figure 8 - Frequency Spectrum referring to the Oceanic Niño Index (a) and the wave parameters after removing seasonality for
the period from Jan 1979 to Dec 2013, being: (b) significant wave height; (c) peak direction and; (d) peak period data. The
red nose spectrum represents a 95% confidence level and the units of frequency are presented in cycles by year (c.b.y.).
Figura 8 - Espectro de frequências referente aos dados de Índice de Niño Oceânico (a) e os parâmetros de onda após remoção da sazonalidade para o período de Janeiro de 1979 a Dezembro de 2013, sendo: (b) onda significativa; (c) direção de
pico e; (d) período de pico. O espectro do “ruído vermelho” representa num nível de confiança de 95% e, as unidades de
frequência são apresentadas em ciclos por ano (c.b.y. – sigla em inglês).
Figure 9 - Cross-correlation between the Oceanic Niño Index and time series for: (a) significant wave height; (b) peak direction; (c) peak period.
Figura 9 - Correlação cruzada entre o Índice de Niño Oceânico e as series temporais para: (a) altura significativa de onda;
(b) direção de pico e; (c) período de pico.
363
Pereira & Klumb-Oliveira (2015)
Figure 10 - Comparison of theoretical and empirical Cumulative Distribution Functions (CDF and ECDF) to a sample population of correlations between Oceanic Niño Index and time series of wave parameters, considering tests with N = 1,000 (a),
(b) and (c) and N = 200 (d), (e) and (f).
Figura 10 - Comparação entre as Funções de Distribuição Cumulativa teóricas (CDF – sigla em inglês) e empíricas (ECDF)
para uma população amostral de correlações entre o Índice de Niño Oceânico e as séries temporais para os parâmetros
de onda, considerando teste com N = 1000 (a), (b) e (c), e N = 200 (d), (e) e (f).
higher concentration of waves from the E-NE. In
autumn, the pattern remained similar to the general
climatology, while in winter a clear reduction in the
percentage of waves from S-SW was noted. These
waves returned to intensify in the spring, exceeding the
percentages present in the general climatology for the
period 1979-2013.
For the El Niño year 1998 (see Supporting Information
IV), which comprises the end of the same El Niño event
referenced above, the wave pattern showed a similar
pattern to that of 1997 throughout the year, except during winter and spring. The winter period showed a high
incidence of waves from the E with low values for the
significant wave height and peak period, while a higher
occurrence of waves from the SE was observed in the
spring.
For the La Niña year 2000 (see Supporting Information
V), a higher percentage of E-NE waves was observed in
summer if compared to the general climatology, as was
observed in El Niño events. A reduction in the occurrence of S-SW waves was observed during the autumn
and an increase in winter. During the spring, an increase
in the percentage of E-NE waves was noted.
Thus, in general, the reduction of the average of significant wave height in El Niño years shown in the crosscorrelation data is seen most clearly with respect to
reducing the percentage of incident waves from the
SSE-SSW during the winter period in those years. The
reverse is true in La Niña years. Also observed in El
Niño events is an increase in the percentage of waves
coming from ENE in summer and autumn, while in the
spring season, La Niña corresponds with an increase of
ENE-ESE waves and decrease in SSE-SSW related to
the El Niño.
Examining the mean pattern of the years’ compositions
with the occurrence of the most persistent El Niño and
La Niña events (Figures 11 and 12, respectively), we
observed a general pattern very similar to that observed
in the seasonal climatology of the 35 years (Figure 6)
for all seasons. It is evident, however, that a slight decrease in SSE-SSW wave percentages occurs during the
winter in El Niño years, while La Niña years have slight
364
Revista de Gestão Costeira Integrada / Journal of Integrated Coastal Zone Management, 15(3):353-370 (2015)
Figure 11 - Directional histogram for wave data for the most persistent El Niño years (1983, 1987, 1992, 1997 and 2003) presented with significant wave height and its period, respectively, for Austral seasons: (a) and (b) - summer (DJF); (c) and (d)
- autumn (MAM); (e) and (f) - winter (JJA); and (g) and (h) - spring (SON).
Figura 11 - Histogramas direcionais dos dados de onda para o os anos de El Niño mais persistente (1983, 1987, 1992, 1997 e
2003) apresentando a altura significativa de onda e seu período de pico, respectivamente, para as estações do ano no hemisfério sul: (a) e (b) – verão (DJF); (c) e (d) – outono (MAM); (e) e (f) – inverno (JJA) e; (g) e (h) – primavera (SON).
365
Pereira & Klumb-Oliveira (2015)
Figure 12 - Directional histogram for wave data for the most persistent La Niña years (1988, 2000 and 2008) presented with
significant wave height and its period, respectively, for Austral seasons: (a) and (b) - summer (DJF); (c) and (d) - autumn
(MAM); (e) and (f) - winter (JJA); and (g) and (h) - spring (SON).
Figura 12 - Histogramas direcionais dos dados de onda para o os anos de La Niña mais persistente (1988, 2000 e 2008) apresentando a altura significativa de onda e seu período de pico, respectivamente, para as estações do ano no hemisfério sul:
(a) e (b) – verão (DJF); (c) e (d) – outono (MAM); (e) e (f) – inverno (JJA) e; (g) e (h) – primavera (SON).
366
Revista de Gestão Costeira Integrada / Journal of Integrated Coastal Zone Management, 15(3):353-370 (2015)
increase in wave percentages from the same direction,
for the same season. This fact corroborates the observations in the analysis of specific El Niño years/La Niña
events, but also shows that these events exhibit different
characteristics through the seasons in different years.
Low correlation values are justified by the small variation in the wave percentages, concentrated in a specific
season. However, the persistence of this characteristic
in various events of El Niño/La Niña needs to be emphasized.
In general, the literature shows that in El Niño events
there is a weakening of the ascending branch of the
Hadley cell over the tropics in South America and the
Intertropical Convergence Zone (ITCZ) is inhibited; the
opposite occurs in La Niña events: an intensification
occurs in ascending and descending branches associated
with Walker and Hadley cells (Souza & Ambrizzi,
2002). Kousky et al. (1984) observe that the strengthening of the Hadley cell favors the intensification of the
trade winds, reflecting a higher transfer of energy to the
E waves in the study area for periods of El Niño, reversing this process in periods of La Niña. In seasonal
climatologies (Figures 6, 11 and 12), it can be seen,
discreetly, that a higher percentage of waves from NESE quadrant occurs in the years of El Niño events.
In El Niño events, there is a reduction in air pressure
over the Pacific, increasing upward movements in the
region, changing the zonal circulation (Walker cell),
which increases the downward movements in tropical
regions such as the Brazilian Northeast. In the state of
Ceará, for example, an intensification of the winds on
the surface is observed in El Niño periods and a weakening of the winds in the La Niña periods (Araújo
Júnior et al., 2014). The atmospheric circulation also
changes in extratropical regions with the intensification
of the trade winds, which causes the obstruction and
therefore a change in trajectory and intensity of frontal
systems (Aragão, 1998).
Using the El Niño event of 1982-1983 as an example,
Kousky et al. (1984) reported both the occurrence of a
significant intensification of upper tropospheric subtropical jets and many blocking events at mid-latitudes.
Combined, these co-occurrences favor the permanence
of persistent frontal systems in southern Brazil, confirmed by Silva et al. (2009). These jets and blocking
events can reduce the approach of these frontal systems
to the region under study. This hypothesis would explain the lower incidence of S-SW waves with high
significant wave height and peak period values during
El Niño years (especially in winter). The south waves
resulting primarily from frontal systems from the southern region should occur with greater frequency in winter. In the other hand, the South Atlantic High weakens
in periods of La Niña, reducing the intensity of the trade
winds and favoring the migration of cold fronts that
affect the southeast.
El Niño periods also influence the coastal sediment
budget through the wave climate. Fernandez & Muehe
(2010) observed that the blocking of cold fronts during
these periods decreased the occurrence of cold fronts on
Rio de Janeiro's coast, with a consequent reduction of
the wave’s transposition events in the area of Massambaba Beach. That reduction inhibited the recovery of
the sedimentary stock, which resulted in a negative
sediment budget for those years.
Despite the variability of the wave percentages from SSW in El Niño/La Niña years being small, we cannot
forget that this is a mean pattern that does not reflect the
exact number of occurrences of cold fronts, their intensity in the study area or their reflections in the wave
pattern. This apparent slight increase in the percentage
of waves in La Niña years, as an example, can strongly
affect the coastal sediment dynamics. A higher average
percentage of waves in the winter as a result of a higher
occurrence of cold fronts in a short period can result in
a high remobilization of sediments along the coastline
without sufficient time for the replacement of these
sediments. This may substantially change the beach
morphology in that period.
Other interannual variability modes can influence the
pattern of incident waves, such as the Tropical Atlantic
Dipole. As for the ENSO in the Pacific, this is the result
of variability in large-scale ocean-atmosphere coupling,
centered on the southern autumn period (Souza & Nobre, 1998). In this case, the sea temperature anomalies
in the surface in the Tropical Atlantic interfere with
meridional circulation (the Hadley cells) and interfere
in the southern shift of the ITCZ. With warmer waters
in the North Atlantic tropical zone and cooler in the
South Atlantic tropical zone, there is an intensification
of the downflow in the southern portion of the tropical
Atlantic; this causes, for example, reduced rainfall in
northeastern Brazil (Aragão, 1998).
It is noteworthy that the action of these two combined
events can promote variability in the responses to atmospheric circulation and consequently to the waves
generated by wind in the South Atlantic, depending on
the separate intensities of these events. This explains
why the variability of wave parameters in an interannual analysis are only partially explained by the ONI
and that the correlation values tend to be reduced.
4. Conclusions
This research aimed to detect and analyze the variability of regional wave climate from Rio de Janeiro state’s
central coast in relation to ENSO events. Wave model
data from WAVEWATCH III (NWW3) available from
NOAA were used to characterize the interannual and
367
Pereira & Klumb-Oliveira (2015)
seasonal wave climate in the region. The model was
validated here with data from an oceanographic buoy
anchored in the study area and a strong correlation observed. The NWW3 data for the Rio de Janeiro continental shelf were also compared to studies in the region
to validate the data.
The correlations between wave parameters and the
Oceanic Niño Index (INO) had many low values, nonetheless significant, that demonstrated the existence of a
relationship between the variabilities of these parameters, particularly for significant wave height. This parameter had a correlation of approximately -0.17, with a
time lag of four months. It represents a small but statistically significant value. This indicated that in El Niño
years, a tendency to reduce the significant wave height
existed with a several month delay in the response of
the local waves, while an inverse pattern occurred in La
Niña years. An increase in wave incidence was observed from the S and SW quadrants in La Niña years,
especially in boreal winter, causing the relative increase
in wave heights values.
The lag observed for significant wave height proved
consistent with values reported for the lag wind, noting
that the variation in waves is to a large extent caused by
the attenuation of the frequency of events from the S
quadrant during El Niño periods. These events are associated with large percentage of the occurrences of frontal systems.
This pattern of reduction in significant wave heights in
El Niño years is noted in the literature, attributed to the
intensification of the center of the South Atlantic High
combined with increase in the occurrence of subtropical
jets in these periods. This results in a larger barrier to
cold fronts, these being retained in the southern region
of Brazil and reaching the Southeast less frequently.
Turning to the low correlation values calculated for the
period and direction peak, these indicate a negligible
interference or a lesser degree of reliability for interpretation as a single variable. However, the combined
analysis of this parameter with the significant wave
height in the directional histograms shows a consistency
in the pattern observed for the different periods analyzed.
The occurrence of other modes of variability with an
interannual character in the Atlantic Ocean influences
the variability of the wave parameters in the regions
where the ONI is responsible for only a portion of this
variability; this increased the plausibility of the low
direct correlation values between the series. A thorough
study of the other modes of interannual variability that
could influence this data will be required.
Therefore, we suggest that analysis of the waves under
the El Niño/La Niña bias should continue, especially in
the southern hemisphere, both in other regions of Brazil
and also along the occidental African coast. This would
enhance the understanding of the remote effects of wind
variability in the Pacific Ocean to other regions of the
Atlantic Ocean and the corresponding response of the
ocean.
Appendix
Supporting Information associated with this article is available online at http://www.aprh.pt/rgci/pdf/rgci-570_Pereira_SupportingInformation.pdf
Acknowledgments
The authors thank the Support Program for Research and Development (Programa de Suporte à Pesquisa e Desenvolvimento - PSPD)
at MAG - Mar, Ambiente e Geologia, the division responsible for
financing, encouragement and providing the physical infrastructure for data processing and scientific development. The authors
also thank the Universidade Federal do Rio de Janeiro (UFRJ) and
the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior
(CAPES) for physical and financial support for the research.
References
Aragão, J.O.R. (1998) - O Impacto do ENSO e do Dipolo do Atlântico no Nordeste do Brasil. Bulletin de l'Institut Français d'Etudes
Andines, 27(3):839–844. DOI: 10.1590/S1415-43662009000400014
Araújo Júnior, l.M.; Silva, E.M.; Costa, A.A.; Sales, D.C.; Vasconcelos Junior, F.C.; Guimarães, S.O. (2014) - Avaliação dos eecursos
eólicos simulado no Estado do Ceará: Um estudo de caso para o El
Niño 97/98 e para La Niña 98/99. Revista Brasileira de Meteorologia, 29(1):139–152. DOI: 10.1590/S0102-77862014000100013
Bastos, A.C.; Silva, C.G. (2000) - Caracterização morfodinâmica do
litoral norte fluminense, RJ, Brasil. Revista Brasileira de Oceanografia, 48(1):41–60. DOI: 10.1590/S1413-77392000000100004
Becker, J.J.; Sandwell, D.T.; Smith, W.H.F.; Braud, J.; Binder, B.;
Depner, J.; Fabre, D.; Factor, J.; Ingalls, S.; Kim, S-H.; Ladner, R.;
Marks, K.; Nelson, S.; Pharaoh, A.; Trimmer, R.; Von Rosenberg,
J.; Wallace, G.; Weatherall, P. (2009) - Global bathymetry and
elevation data at 30 arc seconds resolution: SRTM30_PLUS. Marine Geodesy, 32(4):355–371. DOI: 10.1080/01490410903297766
Caires, S.; Sterl, A.; Bidlot, J.-R.; Graham, N.; Swail, V. (2004) Intercomparison of different wind-wave reanalyses. Journal of
Climate,
17(10):1893–1913.
DOI:
10.1175/15200442(2004)017<1893:IODWR>2.0.CO;2
Campos, R.M. (2009) - Análise dos extremos de onda no Rio de
Janeiro associados a ciclones extratropicais no Atlântico Sul. 241p,
Master's thesis, Universidade Federal do Rio de Janeiro, Rio de
janeiro,
RJ,
Brazil.
Available
on-line
at
http://livros01.livrosgratis.com.br/cp103377.pdf
Changnon, S.A. (1999) Impacts of 1997-98 El Niño Generated
Weather in the United States. Bulletin of the American Meteorological
Society,
80(9):1819–1827.
DOI:
10.1175/15200477(1999)080<1819:IOENOG>2.0.CO;2
Chawla, A.; Spindler, D.; Tolman, H. (2012) - 30 Year Wave Hindcasts using WAVEWATCH III with CFSR winds - Phase I. MMAB
Contribution No. 302: NCEP/NOAA, 12p., College Park, MD,
United
States
of
America.
http://polar.ncep.noaa.gov/mmab/papers/tn302/MMAB_302.pdf
Chawla, A.; Spindler, D.; Tolman, H. L. (2011) - A thirty year wave
hindcast using the latest NCEP Climate Forecast System Reanalysis
winds. Proc. 12th International Workshop on Wave Hindcasting and
Forecasting, Kohala Coast, Hawai’i, HI, p.1-11., College Park, MD,
United
States
of
America.
Available
on-line
at
http://polar.ncep.noaa.gov/mmab/papers/tn296/MMAB_296.pdf
Chen, F.; Shapiro, G.; Thain, R. (2013) - Sensitivity of Sea Surface
Temperature Simulation by an Ocean Model to the Resolution of the
368
Revista de Gestão Costeira Integrada / Journal of Integrated Coastal Zone Management, 15(3):353-370 (2015)
Meteorological Forcing. ISRN Oceanography, 2013(ID 215715):1–
12. DOI: 10.5402/2013/215715
Oceanography,
58(1):53–70.
87592010000100006
Douglass, D.H. (2011) - Separation of a Signal of Interest from a
Seasonal Effect in Geophysical Data: I. El Niño/La Niña Phenomenon. International Journal of Geosciences, 2:414-419.
DOI: 10.4236/ijg.2011.24045
Rogers, W.E.; Dykes, J.D.; Wang, D.; Carroll, S.N.; Watson, K.
(2012) - Validation Test Report for WAVEWATCH III. 67p, DTIC
Document, Slidell, LA, United States of America. Available on-line at
Emery, W.J.; Thomson, R.E. (2001) - Data analysis methods in
physical oceanography. 2nd. ed. 658p., Elsevier, Amsterdam, Netherlands. ISBN: 9780080477008
Enfield, D. B.; Mayer, D. A. (1997) - Tropical Atlantic sea surface
temperature variability and its relation to El Niño Southern Oscillation. Journal of Geophysical Research: Oceans (1978–2012),
102(C1):929–945. DOI: 10.1029/96JC03296
DOI:
10.1590/s1679-
http://oai.dtic.mil/oai/oai?verb=getRecord&metadataPrefix=html&identifier
=ADA576706
Shabbar, A.; Yu, B. (2009) - The 1998–2000 La Niña in the context
of historically strong La Niña events. Journal of Geophysical Research, 114(D13115):1-14. DOI: 10.1029/2008JD011185
Short, A.D. (1999) - Handbook of Beach and Shoreface Morphodynamics. 379p., Wiley, Chichester, United Kingdom. ISBN:
0471965707
Fernandez, G.B.; Muehe, D. (2006) - Algumas considerações sobre
o efeito do fenômeno El Niño sobre feições costeiras ao longo da
costa brasileira. GEOgraphia, 8(15):115-128, Rio de Janeiro, RJ,
Brazil.
Available
on-line
at
Silva, G.A.; Ambrizzi, T.; Marengo, J.A. (2009) - Observational
evidences on the modulation of the South American Low Level Jet
east of the Andes according the ENSO variability. Annales geophysicae, 27:645-657. DOI: 10.5194/angeo-27-645-2009
http://www.uff.br/geographia/ojs/index.php/geographia/article/viewArticle/191
Smith, T.M.; Reynolds, R.W.; Peterson, T.C.; Lawrimore, J. (2008)
- Improvements to NOAA’s Historical Merged Land–Ocean Surface
Temperature Analysis (1880–2006). Journal of Climate,
21(10):2283–2296. DOI: 10.1175/2007JCLI2100.1
Hanson, J.L.; Tracy, B.A.; Tolman, H.L.; Scott, R.D. (2009) - Pacific Hindcast Performance of Three Numerical Wave Models.
Journal of Atmospheric and Oceanic Technology, 26(8):1614–1633.
DOI: 10.1175/2009JTECHO650.1
Hetland, R.D.; Dimarco, S.F. (2012) - Skill assessment of a hydrodynamic model of circulation over the Texas–Louisiana continental
shelf.
Ocean
Modelling,
43-44:64–76.
DOI:
10.1016/j.ocemod.2011.11.009
Holthuijsen, L.H. (2007) - Waves in oceanic and coastal waters.
387p., Cambridge University Press, Cambridge, United Kingdom.
ISBN: 9780511618536
Kaplan, E.L.; Meier, P. (1958) - Nonparametric Estimation from
Incomplete Observations. Journal of the American Statistical Association, 53(282):457-481. DOI: 10.2307/2281868
Komar, P.D. (1976) - Beach processes and sedimentation. 429p.,
Prentice-Hall, Inc., New Jersey, United States of America. ISBN:
0130725951.
Kousky, V.E.; Kagano, M.T.; Cavalcanti, I.F. (1984) - A review of
the Southern Oscillation: oceanic‐atmospheric circulation changes
and related rainfall anomalies. Tellus, 36(5):490–504. DOI:
10.1111/j.1600-0870.1984.tb00264.x
Laing, A. K. (1998) - An Introduction to Ocean Waves. In: E.
Bouws (org.), Guide to Wave analysis and forecasting, pp.1–14, 2nd
Ed., World Meteorological Organization, Geneva, Switzerland.
Available
on-line
at
ISBN:
9263127026.
http://www.wmo.int/pages/prog/amp/mmop/documents/WMO%20No%2070
2/WMO702.pdf
Larkin, N.K.; Harrison, D.E. (2005) - On the definition of El Niño
and associated seasonal average U.S. weather anomalies. Geophysical
Research
Letters,
31(L12705):1-4.
DOI:
10.1029/2005GL022738
Lawless, J.F. (2002) - Statistical models and methods for lifetime
data. 2nd Edition, 664p., Wiley-Interscience, John Wiley & Sons,
Hoboken, NJ, USA. ISBN: 9780471372158
Muehe, D.; Corrêa, C.H.T. (1989) - Dinâmica de praia e transporte
de sedimentos na restinga de Maçambaba, RJ. Revista Brasileira de
Geociências (ISSN: 0375-7536), 19(3):387–392, Sociedade Brasileira de Geologia, São Paulo, SP, Brazil.
Muehe, D. (2011) - Erosão Costeira - Tendência ou Eventos Extremos? O Litoral entre Rio de Janeiro e Cabo Frio, Brasil. Revista de
Gestão Costeira Integrada, 11(3):315-325. DOI: 10.5894/rgci282
Orcutt, G.H.; James, S.F. (1948) - Testing the significance of correlation between time series. Biometrika, 35(3/4):397–413. Article
Stable URL: http://www.jstor.org/stable/2332358
Pianca, C.; Mazzini, P.L.F.; Siegle, E. (2010) - Brazilian offshore
wave climate based on NWW3 reanalysis. Brazilian Journal of
Souza, E. B.; Nobre, P. (1998) - Uma revisão sobre o Padrão de
Dipolo no Oceano Atlântico tropical. Revista Brasileira de MeteoroAvailable
on-line
at
logia,
13(1):31–44.
http://www.rbmet.org.br/port/revista/revista_dl.php?id_artigo=455&id_arqui
vo=715.
Souza, E.B.; Ambrizzi, T. (2002) - ENSO impacts on the South
American rainfall during 1980s: Hadley and Walker circulation.
Atmósfera (ISSN: 0187-6236) 15:105–120. Universidad Nacional
Autónoma de México, D.F., México. Available on-line at
http://www.ejournal.unam.mx/atm/Vol15-2/ATM15203.pdf
Souza, M.H.S.; Ribeiro, C.E.P. (1988) - Wave Climate off Rio de
Janeiro. Coastal Engineering Proceedings, 1(21):261–269. DOI:
10.9753/icce.v21.
Spindler, D.M.; Chawla, A.; Tolman, H.L. (2011) - An initial look at
the CFSR Reanalysis winds for wave modeling. MMAB Contribution No. 290: NCEP/NOAA, 23p. College Park, MD, U.S.A. Disponível em: http://polar.ncep.noaa.gov/mmab/papers/tn290/MMAB_290.pdf
Tolman, H.L. (2002) - User manual and system documentation of
WAVEWATCH III version 2.22. MMAB Contribution No. 222,
NCEP/NOAA, 133p. College Park, MD, U.S.A. Available on-line at
http://polar.ncep.noaa.gov/mmab/papers/tn222/MMAB_222.pdf
Tolman, H.L. (2009) - User manual and system documentation of
WAVEWATCH III version 3.14. MMAB Contribution No. 276,
NCEP/NOAA, 194p. College Park, MD, U.S.A. Available on-line at
http://polar.ncep.noaa.gov/mmab/papers/tn276/MMAB_276.pdf
Trenberth, K.E. (1997) - The Definition of El Niño. Bulletin of the
American Meteorological Society, 78(12):2771–2777. DOI:
10.1175/1520-0477(1997)078<2771:TDOENO>2.0.CO;2
Trenberth, K.E.; Stepaniak, D. P. (2001) - Indices of El Niño evolution. Journal of Climate, 14(8):1697–1701. DOI: 10.1175/15200442(2001)014<1697:LIOENO>2.0.CO;2
Vale, E.M.J.P. (2012) - Analise de situações meteo-oceanográficas
severas de bom tempo na Bacia de Campos. 80p, Master's thesis,
Universidade Federal do Rio de Janeiro, Rio de Janeiro, RJ, Brazil.
Available
on-line
at
http://www.oceanica.ufrj.br/intranet/modules/PDdownloads/singlefile.php?ci
d=40&lid=759
Willmott, C.J. (1981) - On the validation of models. Physical geography, 2(2):184–194. DOI: 10.1080/02723646.1981.10642213
Willmott, C.J. (1982) - Some comments on the evaluation of model
performance. Bulletin of the American Meteorological Society,
63(11):1309–1313.
DOI:
10.1175/15200477(1982)063<1309:SCOTEO>2.0.CO;2
Willmott, C.J.; Matsuura, K. (2005) - Advantages of the mean absolute error (MAE) over the root mean square error (RMSE) in assess-
369
Pereira & Klumb-Oliveira (2015)
ing average model performance. Climate Research (ISSN: 16161572), 30(1):79-82, Inter-Research Science Center. Available on-line
at http://climate.geog.udel.edu/~climate/publication_html/Pdf/WM_CR_05.pdf
Zang, Z.; Moore, J. (2011) - New significance test methods for
Fourier analysis of geophysical time series. Nonlinear Processes in
Geophyics, 18:643–652. DOI: 10.5194/npg-18-643-2011
Wolter, K.; Timlin, M.S. (1998) - Measuring the strength of ENSO
events: how does 1997/98 rank? Weather, 53(9):315–324.
DOI: 10.1002/j.1477-8696.1998.tb06408.x
370